Many real-world datasets, such as citation networks, social networks, and molecular structures, are naturally represented as heterogeneous graphs, where nodes belong to different types and have additional features. For example, in a citation network, nodes representing ''Paper'' or ''Author'' may include attributes like keywords or affiliations. A critical machine learning task on these graphs is node classification, which is useful for applications such as fake news detection, corporate risk assessment, and molecular property prediction. Although Heterogeneous Graph Neural Networks (HGNNs) perform well in these contexts, their predictions remain opaque. Existing post-hoc explanation methods lack support for actual node features beyond one-hot encoding of node type and often fail to generate realistic, faithful explanations. To address these gaps, we propose DiGNNExplainer, a model-level explanation approach that synthesizes heterogeneous graphs with realistic node features via discrete denoising diffusion. In particular, we generate realistic discrete features (e.g., bag-of-words features) using diffusion models within a discrete space, whereas previous approaches are limited to continuous spaces. We evaluate our approach on multiple datasets and show that DiGNNExplainer produces explanations that are realistic and faithful to the model's decision-making, outperforming state-of-the-art methods.

Although relational graph neural networks (RGNNs) excel at learning from graph-structured data as it appears in knowledge graphs, they often lack interpretability. While natural language (NL) explanations offer a promising solution, evaluating these explanations remains largely unaddressed. There is no unified evaluation framework including evaluation metrics, benchmarking datasets, and established evaluation procedures. This paper introduces NLEF, a novel NL Evaluation Framework to assess the quality of NL explanations for RGNNs. It uses the NL explanations to make new predictions and assesses in how far they align with the predictions produced by the RGNN. Towards this end, we propose two methods: (1) we convert NL explanations to description logics (DL) and use a DL reasoner for node classifications, (2) we use a retrieval augmented approach (RAG) for node classifications. Our evaluation results show that our DL method is highly scalable, whereas the RAG approach often yields the highest performance.

Concept learning exploits background knowledge in the form of description logic axioms to learn explainable classification models from knowledge bases. Despite recent breakthroughs in neuro-symbolic concept learning, most approaches still cannot be deployed on real-world knowledge bases. This is due to their use of description logic reasoners, which are not robust against inconsistencies nor erroneous data. We address this challenge by presenting a novel neural reasoner dubbed Ebr. Our reasoner relies on embeddings to approximate the results of a symbolic reasoner. We show that Ebr solely requires retrieving instances for atomic concepts and existential restrictions to retrieve or approximate the set of instances of any concept in the description logic \(\mathcal {SHOIQ}\). In our experiments, we compare Ebr with state-of-the-art reasoners. Our results suggest that Ebr is robust against missing and erroneous data in contrast to existing reasoners.

As machine learning and deep learning become increasingly integrated into our daily lives, understanding how these technologies make decisions is crucial. To ensure transparency, accountability, and ethical adherence, these so-called "black-box" models should be accompanied by human-comprehensible explanations of their predictions. This clarity is essential for establishing trust in their real-world applications. Similarly, it is crucial to compare different types of explanations to evaluate and understand their effectiveness, interpretability, and generalization capabilities for informed selection in various applications. To this end, we propose a framework called EDGE to evaluate diverse knowledge graph explanations, assessing logical rule-based and subgraph-based explanations by various explainers in terms of prediction accuracy and fidelity to the Graph Neural Network (GNN) model. Our evaluations reveal that logical methods excel in explaining complex and structured data, while subgraph-based models exhibit higher fidelity to the GNN model, earning them the label "GNN Explainers". Although further diversified evaluations are necessary to determine the superiority of one explanation type over another, our study shows that each type has pros and cons.

Causal questions inquire about causal relationships between different events or phenomena. They are important for a variety of use cases, including virtual assistants and search engines. However, many current approaches to causal question answering cannot provide explanations or evidence for their answers. Hence, in this paper, we aim to answer causal questions with a causality graph, a large-scale dataset of causal relations between noun phrases along with the relations' provenance data. Inspired by recent, successful applications of reinforcement learning to knowledge graph tasks, such as link prediction and fact-checking, we explore the application of reinforcement learning on a causality graph for causal question answering. We introduce an Actor-Critic-based agent which learns to search through the graph to answer causal questions. We bootstrap the agent with a supervised learning procedure to deal with large action spaces and sparse rewards. Our evaluation shows that the agent successfully prunes the search space to answer binary causal questions by visiting less than 30 nodes per question compared to over 3,000 nodes by a naive breadth-first search. Our ablation study indicates that our supervised learning strategy provides a strong foundation upon which our reinforcement learning agent improves. The paths returned by our agent explain the mechanisms by which a cause produces an effect. Moreover, for each edge on a path, our causality graph provides its original source allowing for easy verification of paths.

Node classification is an important task in many fields, e.g., predicting entity types in knowledge graphs, classifying papers in citation graphs, or classifying nodes in social networks. In many cases, it is crucial to explain why certain predictions are made. Towards this end, concept learning has been proposed as a means of interpretable node classification: given positive and negative examples in a knowledge base, concepts in description logics are learned that serve as classification models. However, state-of-the-art concept learners, including EvoLearner and CELOE exhibit long runtimes. In this paper, we propose to accelerate concept learning with graph sampling techniques. We experiment with seven techniques and tailor them to the setting of concept learning. In our experiments, we achieve a reduction in training size by over 90% while maintaining a high predictive performance.

Classifying nodes in knowledge graphs is an important task, e.g., for predicting missing types of entities, predicting which molecules cause cancer, or predicting which drugs are promising treatment candidates. While black-box models often achieve high predictive performance, they are only post-hoc and locally explainable and do not allow the learned model to be easily enriched with domain knowledge. Towards this end, learning description logic concepts from positive and negative examples has been proposed. However, learning such concepts often takes a long time and state-of-the-art approaches provide limited support for literal data values, although they are crucial for many applications. In this paper, we propose EvoLearner—an evolutionary approach to learn concepts in , which is the attributive language with complement () paired with qualified cardinality restrictions () and data properties (). We contribute a novel initialization method for the initial population: starting from positive examples, we perform biased random walks and translate them to description logic concepts. Moreover, we improve support for data properties by maximizing information gain when deciding where to split the data. We show that our approach significantly outperforms the state of the art on the benchmarking framework SML-Bench for structured machine learning. Our ablation study confirms that this is due to our novel initialization method and support for data properties.

Entity typing in knowledge graphs (KGs) aims to infer missing types of entities and might be considered one of the most significant tasks of knowledge graph construction since type information is highly relevant for querying, quality assurance, and KG applications. While supervised learning approaches for entity typing have been proposed, they require large amounts of (manually) labeled data, which can be expensive to obtain. In this paper, we propose a novel approach for KG entity typing that leverages semi-supervised learning from massive unlabeled data. Our approach follows a teacher-student paradigm that allows combining a small amount of labeled data with a large amount of unlabeled data to boost performance. We conduct several experiments on two benchmarking datasets (FB15k-ET and YAGO43k-ET). Our results demonstrate the effectiveness of our approach in improving entity typing in KGs. Given type information for only 1% of entities, our approach ASSET predicts missing types with a F1-score of 0.47 and 0.64 on the datasets FB15k-ET and YAGO43k-ET, respectively, outperforming supervised baselines.

Causal knowledge is seen as one of the key ingredients to advance artificial intelligence. Yet, few knowledge bases comprise causal knowledge to date, possibly due to significant efforts required for validation. Notwithstanding this challenge, we compile CauseNet, a large-scale knowledge base of claimed causal relations between causal concepts. By extraction from different semi- and unstructured web sources, we collect more than 11 million causal relations with an estimated extraction precision of 83% and construct the first large-scale and open-domain causality graph. We analyze the graph to gain insights about causal beliefs expressed on the web and we demonstrate its benefits in basic causal question answering. Future work may use the graph for causal reasoning, computational argumentation, multi-hop question answering, and more.

Crowdsourced knowledge bases like Wikidata suffer from low-quality edits and vandalism, employing machine learning-based approaches to detect both kinds of damage. We reveal that state-of-the-art detection approaches discriminate anonymous and new users: benign edits from these users receive much higher vandalism scores than benign edits from older ones, causing newcomers to abandon the project prematurely. We address this problem for the first time by analyzing and measuring the sources of bias, and by developing a new vandalism detection model that avoids them. Our model FAIR-S reduces the bias ratio of the state-of-the-art vandalism detector WDVD from 310.7 to only 11.9 while maintaining high predictive performance at 0.963 ROC and 0.316 PR.

The WSDM Cup 2017 was a data mining challenge held in conjunction with the 10th International Conference on Web Search and Data Mining (WSDM). It addressed key challenges of knowledge bases today: quality assurance and entity search. For quality assurance, we tackle the task of vandalism detection, based on a dataset of more than 82 million user-contributed revisions of the Wikidata knowledge base, all of which annotated with regard to whether or not they are vandalism. For entity search, we tackle the task of triple scoring, using a dataset that comprises relevance scores for triples from type-like relations including occupation and country of citizenship, based on about 10,000 human relevance judgments. For reproducibility sake, participants were asked to submit their software on TIRA, a cloud-based evaluation platform, and they were incentivized to share their approaches open source.

Wikidata is the new, large-scale knowledge base of the Wikimedia Foundation. Its knowledge is increasingly used within Wikipedia itself and various other kinds of information systems, imposing high demands on its integrity. Wikidata can be edited by anyone and, unfortunately, it frequently gets vandalized, exposing all information systems using it to the risk of spreading vandalized and falsified information. In this paper, we present a new machine learning-based approach to detect vandalism in Wikidata. We propose a set of 47 features that exploit both content and context information, and we report on 4 classifiers of increasing effectiveness tailored to this learning task. Our approach is evaluated on the recently published Wikidata Vandalism Corpus WDVC-2015 and it achieves an area under curve value of the receiver operating characteristic, ROC-AUC, of 0.991. It significantly outperforms the state of the art represented by the rule-based Wikidata Abuse Filter (0.865 ROC-AUC) and a prototypical vandalism detector recently introduced by Wikimedia within the Objective Revision Evaluation Service (0.859 ROC-AUC).

We report on the construction of the Wikidata Vandalism Corpus WDVC-2015, the first corpus for vandalism in knowledge bases. Our corpus is based on the entire revision history of Wikidata, the knowledge base underlying Wikipedia. Among Wikidata's 24 million manual revisions, we have identified more than 100,000 cases of vandalism. An in-depth corpus analysis lays the groundwork for research and development on automatic vandalism detection in public knowledge bases. Our analysis shows that 58% of the vandalism revisions can be found in the textual portions of Wikidata, and the remainder in structural content, e.g., subject-predicate-object triples. Moreover, we find that some vandals also target Wikidata content whose manipulation may impact content displayed on Wikipedia, revealing potential vulnerabilities. Given today's importance of knowledge bases for information systems, this shows that public knowledge bases must be used with caution.